all_results.json

{
    "epoch": 1.0,
    "eval_logits/chosen": -7.432624816894531,
    "eval_logits/rejected": -7.451437473297119,
    "eval_logps/chosen": -606.1920166015625,
    "eval_logps/rejected": -550.2219848632812,
    "eval_loss": 0.6320449113845825,
    "eval_nll_loss": 2.2297186851501465,
    "eval_rewards/accuracies": 0.640999972820282,
    "eval_rewards/chosen": 3.7593398094177246,
    "eval_rewards/margins": 0.7855273485183716,
    "eval_rewards/rejected": 2.9737422466278076,
    "eval_runtime": 32.3847,
    "eval_samples_per_second": 30.879,
    "eval_steps_per_second": 7.72
}